It is oftentimes necessary that tunable elements such as cavities or resonant structures coupled to a reference line of interest be frequency stabilized, at least to some degree, so that frequency of the output signals derived from the reference line are dependable for utilization purposes. Alternately, it is oftentimes desireable to stabilize the frequency of length of a cavity to a reference resonance line.
Where the device is to operate as a frequency standard, frequency stability is vital and such stability must be made to extend over a long term for best utilization of such a standard. Likewise, where the device is to operate as a length standard, length stability must be made to extend over as long a time as possible.
Frequency standards, clocks, and length standards of many configurations have been heretofore proposed and/or utilized based on atomic, molecular, or other resonance lines. Such devices have heretofore normally depended, for stability, upon the stability of the components, and this was particularly true where resonant cavities were utilized. Such cavities have commonly heretofore been made stable by using stable materials such as fused quartz, Cervit, ULE quartz, etc.
The long term stability of such devices, however, at least in some configurations, has been limited by a plurality of factors one of the most important being cavity pulling. Cavity pulling is due to the radiation used to probe the atomic, molecular, or other reference resonance line of such a device also being coupled to other resonant structures (typically cavities of some sort) which can change the apparent center of the reference resonance so that the output signal is perturbed (pulled) by the detuning of the cavity. In the case where the atomic, molecular or other resonance line is observed by sampling the same radiation that probes the resonance, then the shift of the apparent center of the resonance is on the order of EQU (.DELTA..nu./.nu.) = (Q.sub.c /Q.sub.R) (.nu..sub.R - .nu..sub.c)
where Q.sub.c is the quality factor of the cavity (resonant frequency divided full width at one-half power), Q.sub.R is the quality factor of the atomic or molecular resonance of interest, and (.nu..sub.R - .nu..sub.c) is the detuning of the cavity from the atomic or molecular resonance.
For a hydrogen maser frequency standard, Q.sub.R is the quality factor of the hydrogen resonance (1,0 to 0,0 transition in the ground state of atomic hydrogen) with (.nu..sub.R - .nu..sub.c) being the detuning of the cavity resonant frequency from the hydrogen resonance. In order to achieve a stability of 10.sup.-14, .DELTA..nu. must be smaller than 1.4 .times. 10.sup.-5 Hz. Present hydrogen masers typically have a cavity quality factor on the order of 30,000 and a hydrogen line Q.sub.R of 10.sup.9. Limits of about these values are imposed by the need to sustain oscillation. This requires a cavity stability of 0.5 Hz or a fractional stability of 3 .times. 10.sup.-10 for a mechanical cavity nearly 30 centimeters in diameter and 30 centimeters tall. In addition to its mechanical size, the electrical properties of the cavity depend on the coupling loops to external amplifiers and surface conditions. Any of the previous factors are more than sufficient to fractionally change the cavity resonance by more than 3 .times. 10.sup.-10 in long-term.
The present success in using the hydrogen maser as a stable frequency standard is directly traceable to the use of spin exchange collisions between hydrogen atoms to compensate for the cavity pulling (see e.g. S. B. Crampton Ph.D thesis Harvard University 1964 and also U.S. Pat. No. 3,792,368). Briefly, the density of atomic hydrogen is modulated and the cavity tuned until no frequency or alternately no phase change occurs synchronously with the hydrogen modulation. This presently requires a minimum of two hydrogen masers to be even partly successful and experimentally still has not proved able to produce stabilities below .about.10.sup.-14 for measurement times greater than a few days even though attempts have been made by a number of groups over a period of 10 years.
The cavity pulling effect also applies to devices which do not sample the radiation responsible for causing the atomic, molecular or other resonance transition of interest. In devices like Cesium frequency standards, for example, the detection is based on some secondary process which is proportional to the number of atoms undergoing the appropriate transition. In this case, the cavity pulling is on the order of EQU .DELTA..nu. = (Q.sub.c /Q.sub.R).sup.2 (.nu..sub.R - .nu..sub.c)
and typically is much less important. However, in some special cases it is still necessary to stabilize the resonance cavity in which case this invention could be used.
An atomic clock or frequency standard utilizing a source of atomic hydrogen in conjunction with a tuned cavity and local oscillator is shown, for example, in U.S. Pat. No. 3,792,368. In this patent, a device and method are taught for tuning the resonant frequency of the microwave cavity of a maser oscillator to approximately the transition frequency of the stimulated emission of the active medium of the maser. In this method the resonant frequency of the microwave cavity is corrected using the error signal obtained by synchronously detecting the phase modulation of the maser oscillator caused by modulation of the oscillation amplitude. Multiple modulation techniques, however, are not utilized for achieving frequency stability, neither is the cavity detuning detected by inserting a phase modulated probe frequency.
Hydrogen frequency standards, whether active or passive, are based on the F = 1, mf = 0 to F = 0, mf = 0 hyperfine transition at 1420 MHz in the ground state of atomic hydrogen. In the typical active type of hydrogen maser, wherein no microwave signal is injected into the cavity, various parameters are adjusted, (such as hydrogen beam intensity, storage time, cavity Q, etc.) so that the energy radiated by the hydrogen atoms can be made to exceed cavity losses, and the system breaks into oscillation. The weak signal produced (about 10.sup.-12 to 10.sup.-14 W) is then phase compared with a local oscillator, using mulitplication and hetrodyne techniques in order to preserve signal-to-noise. The output of the phase comparator is then used to phase-lock the local oscillator to the hydrogen signal.
Several passive hydrogen frequency standards have also been heretofore suggested. The first known passive frequency standard was proposed and built by H. Hellwig in 1970 [see H. Hellwig, Metrologis 6,56 (1970) and H. Hellwig, H, Bell Metrologia 8,96, (1972)]. In this type of hydrogen frequency standard, a microwave signal is injected into the cavity region. Phase comparison of the output signal with the input signal allows frequency locking of the local oscillator to the hydrogen phase dispersion signal. In the system utilized, the hydrogen signal was separated from other dispersive effects by square wave modulating the H signal at .about.1 Hz via hydrogen beam modulation or Zeeman quenching and detecting the resulting phase modulation with a 1Hz synchronous detector. A frequency stability of 2 .times. 10.sup.-12.sbsp..tau..sup.-1/2 in a 30 Hz bandwidth for measurement times from 1 to 100 s were obtained using this system.
For timekeeping applications, at least one main goal is to minimize the time dispersion over many days or even years if possible. At present, the best that has been accomplished with any frequency standard, including hydrogen and cesium standards, is several parts in 10.sup.-13 per year with the only possible exceptions being the primary cesium standards. Requirements in navigation, such as for global positioning systems, are roughly equivalent to a frequency stability of 1 .times. 10.sup.-14 for 10 days. Thus, systems and methods for achieving even better long-term stability are still desirable, particularly where such systems and methods can offer more stability, be of lower cost, and/or involve structures of reduced volume and/or weight.
In addition, systems and methods are also desirable that have the capability of substantially eliminating cavity pulling, with such systems and methods being applicable to frequency standards and also to other applications, including external cell stabilized lasers, where stabilization of cavity resonance frequency which is determined by its length is important, and also including devices which use the methods of frequency control for the purpose of metrology such as a magnetometer whose sensor is a resonance line whose frequency is proportional to the magnetic field, etc.